Impact of Immunomodulating Therapy on Morbidity in ... - ATS Journals

Impact of Immunomodulating Therapy on Morbidity in Patients with Severe Sepsis DIDIER PITTET, STEPHAN HARBARTH, PETER M. SUTER, KONRAD REINHART, ANTON LEIGHTON, CHRISTOPHER BARKER, FRANCES MACDONALD, EDWARD ABRAHAM, and the Ro 45-2081 Study Group Infection Control Program and Division of Surgical Intensive Care, Geneva University Hospitals, Geneva, Switzerland; Department of Anesthesia and Intensive Care, University Hospital of Jena, Germany; F. Hoffmann-La Roche Ltd., Basel, Switzerland; and Division of Critical Care Medicine, University of Colorado Health Sciences Center, Denver, Colorado

We assessed the impact, over a 28-d period, of therapy with the tumor necrosis factor (TNF) neutralizing receptor fusion protein (p55-IgG) on the incidence of end-organ failures in patients with severe sepsis or early septic shock in a subgroup of 165 patients recruited into a randomized, multicenter clinical trial to receive placebo (n 5 78) or a single infusion of p55-IgG, 0.083 mg/kg (n 5 87). At study entry, distribution of organ dysfunctions and other baseline characteristics were similar for the two study groups. Treatment with p55-IgG was associated with a trend toward reduced 28-d mortality (p 5 0.07), a decreased incidence of new organ dysfunctions (relative risk [RR], 0.57; 95% confidence interval [95% CI] 0.29 to 1.10, p 5 0.10), and a decreased overall incidence-density of organ failures (RR 0.65; 95% CI 0.60 to 0.71, p 5 0.0001). Patients treated with p55-IgG had more organ failure–free days after study entry than those who received placebo. Average intensive care unit (ICU) stay was 2.6 d shorter (95% CI 0.2 to 5.0) for patients who received p55-IgG than for those who received placebo. For those patients who survived, this difference was 4.1 d (95% CI 1.6 to 6.6). Duration of ventilatory support was 3.2 d shorter (95% CI 0.1 to 6.3) among 28-d survivors who received p55-IgG, compared with placebo. In conclusion, in the population of septic patients studied, treatment with p55-IgG was associated with a trend toward shorter need for mechanical ventilatory support, a decreased length of stay (LOS), and a decreased incidence and duration of organ failure. Pittet D, Harbarth S, Suter PM, Reinhart K, Leighton A, Barker C, Macdonald F, Abraham E, and the Ro 45-2081 Study Group. Impact of immunomodulating therapy on morbidity in patients with severe sepsis. AM J RESPIR CRIT CARE MED 1999;160:852–857.

Sepsis has a major impact on the outcome of critically ill patients. The development of circulatory shock, acute respiratory distress syndrome (ARDS), and multiple organ dysfunction syndrome is observed in 40 to 50% of patients with sepsis, and is associated with an increased mortality (1–4). Outcome is related to the number and duration of organ dysfunctions evolving during the course of sepsis (5). The mortality attributable to nosocomial sepsis in critically ill patients was estimated to be 35% and such events were associated with signifi-

(Received in original form September 7, 1998 and in revised form February 10, 1999 ) No part of this work has been funded by tobacco industry sources. Presented in part at the 37th Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC), Toronto, Canada, Sept. 28–Oct. 1, 1997, Abstract #G35. The original study was supported by F. Hoffmann-La Roche Ltd., Basel, Switzerland. Dr. Harbarth’s current affiliation is the Department of Epidemiology, Harvard School of Public Health. Lenercept is produced by Genentech Inc., South San Francisco, CA. Correspondence and requests for reprints should be addressed to Didier Pittet, M.D., M.S., Infection Control Program, Department of Internal Medicine, University of Geneva Hospitals, 1211 Geneva 14, Switzerland. E-mail: didier.pittet@ hcuge.ch Am J Respir Crit Care Med Vol 160. pp 852–857, 1999 Internet address: www.atsjournals.org

cantly increased length of intensive care unit (ICU) and hospital stay, resulting in extra costs of US $40,000 per survivor (6). Tumor necrosis factor (TNF) holds a key position within the network of mediators involved in the immune response during sepsis (7, 8). Serum levels of TNF are elevated in patients with gram-negative sepsis (9) and were directly related to the outcome in patients with meningococcal disease and cerebral malaria (10–12). Different approaches to neutralize TNF have been tested in both animal models (13–15) and clinical trials yielding variable results so far (16–20). Nevertheless, effective immunotherapies may prevent the development of organ dysfunctions (13, 15, 18, 21–23) and thereby improve outcome (13, 14, 20–22). In contrast, the mean length of stay (LOS) in the ICU has previously been observed to be longer for survivors than nonsurvivors, and improved outcome was associated with prolonged ICU stay (6, 24, 25). In the present study of a predetermined subgroup of septic patients recruited into a large randomized clinical trial, we explored the impact of immunotherapy such as TNF neutralizing receptor fusion protein on the incidence of end-organ dysfunctions and the associated use of supportive therapies. In particular, we challenged the hypothesis that a decreased 28-d mortality is associated with a prolonged duration of intensive care to support failing organs. Thus, effective treatment may increase resource use unless such treatment also has a favorable impact on patient morbid-

853

Pittet, Harbarth, Suter, et al.: Immunomodulating Therapy and Sepsis Morbidity

ity, which is considered by some experts as a key endpoint for future clinical trials in intensive care (26).

METHODS Design Patients (. 18 yr of age) were randomized in a Phase II, double-blind, placebo-controlled, multicenter trial to receive a single dose of either 0.008 mg/kg (1 mg/120 kg), 0.042 mg/kg (5 mg/120 kg), or 0.083 mg/kg (10 mg/120 kg) of p55-TNF receptor fusion protein (p55-IgG, lenercept) or placebo. Separate randomization lists were used for the patients stratified into the late (refractory) shock and severe sepsis groups (20). Because a planned interim analysis showed no evidence of benefit in the group of patients with 0.008 mg/kg arm of the study, this dosage was discontinued, and the study was then continued with three arms (placebo, lenercept 0.042 mg/kg, and lenercept 0.083 mg/ kg) until 498 patients were enrolled (20). Patients were prospectively stratified into severe sepsis (with or without early septic shock, n 5 247) and late septic shock (n 5 251) groups, using inclusion criteria as previously described (20). In brief, all patients with severe sepsis had at least two criteria indicating inadequate organ function, but did not fulfill predetermined criteria for late shock (i.e., of more than 12 h duration) prior to study entry. Patients prospectively stratified into the late shock group at study entry were not included in the present analysis.

Organ Failures The definitions of inadequate organ function (20) were used as entry criteria for the study. However, for the purpose of the study analyses the definitions of Hebert and coworkers (27) were used, as these describe, organ failures of significantly greater severity than inadequate organ functions as used for study entry. ARDS and non-ARDS pulmonary failure were considered separately (20). In brief, organ failures assessed at baseline and over the 28 d after study entry were characterized as follows: (1) renal: creatinine greater than 3.0 mg/dl or need for acute dialysis in the absence of chronic renal insufficiency; in the case of preexisting renal dysfunction, a doubling of previous serum creatinine values; (2) hepatic: acute elevation of total bilirubin to greater than 3.0 mg/dl and elevation of alanine aminotransferase (ALT) and/or aspartate transaminase (AST) to three times upper limit of normal in the absence of primary liver disease; (3) hematologic: one or more of the following: peripheral white blood cell count , 2.0 3 109/L, platelet count , 40 3 109/L, or disseminated intravascular coagulation (DIC) defined as prothrombin time (PT) . 23 normal range, partial thromboplastin time (PTT) . 23 normal range, platelet count , 100 3 109/L or less than one-half of a previous count, and fibrin degradation products . 10 mg/L (or D-dimers . 500 mg/L); (4) central nervous system: Glasgow coma scale , 10 or a decrease in the scale of at least 3 if primary central nervous system (CNS) injury is present; (5) ARDS: bilateral pulmonary infiltrate on chest radiograph and an alveolar–arterial (A-a) gradient . 250 mm Hg or ratio of arte-

rial oxygen tension to fraction of inspired oxygen (PaO2/FIO2) less than 280 in the absence of congestive heart failure or other lung disease (PaO2/FIO2 , 200 in the presence of pneumonia); (6) non-ARDS pulmonary failure: respiratory rate < 5 breaths/min or > 50 breaths/min, or requirement for mechanical ventilation without fulfillment of ARDS criteria as defined previously.

Study Variables and Outcome Measures Variables recorded at baseline values and during the 28-d period after study entry included vital signs, physical examinations, data necessary for computation of an Acute Physiology and Chronic Health Evaluation (APACHE) III sepsis severity score (28), routine blood tests, blood and other microbiological cultures, chest X-rays, and measurement of end-organ dysfunctions. Baseline predicted mortality for each patient was calculated using a sepsis mortality risk prediction model provided by APACHE Medical Systems Inc. that was customized to the study protocol’s inclusion and exclusion criteria (28). Outcome measures were: 28-d all-cause mortality; LOS in the ICU stay after study entry; use and duration of intubation for mechanical ventilatory support; number and duration of organ failures associated with severe sepsis that developed after study entry. Number of organ failure–free days (days without organ system failure) were calculated for clinically important organ systems during the 28 d after study entry as previously reported by Bernard and colleagues (23). In brief, a patient who survived 28 d and had no ARDS was assigned a score of 28; a patient who had ARDS for 10 d and died on Day 15, was assigned a score of 5 (5 ARDS-free days). Calculations were made for each organ separately, and then summed if applicable.

Statistical Analysis Rates of end-organ dysfunction were calculated as cases per 100 patients and as cases per 1,000 patient-days. The incidence-density was calculated as the number of days when the patient met the criteria for the outcome of interest (end-organ dysfunction) per 1,000 patientdays. Incidence-density ratios were expressed as relative risks (RR). Differences in proportions were compared using either the chi-square test or Fisher exact test for expected cell frequencies less than 5. Yates correction was used when applicable. LOS and the duration of mechanical ventilatory support were estimated using the Kaplan-Meier estimate of the mean (29). Patients who remained in the ICU at the end of the 28-d study were included in the analysis and their duration of stay was censored at this point. Duration of mechanical ventilatory support was measured as the number of days until first extubation. For the purpose of this analysis, a patient was considered extubated if 3 consecutive days elapsed without mechanical ventilatory assistance or intubation. Differences between treatment means for these three measurements were summarized using 95% confidence intervals (95% CI). In order to account for differences in mortality between the treatment groups, the estimations of mean duration of stay or of ventilatory support among patients discharged alive were computed using a competing risk meth-

TABLE 1 PRESENCE OF ORGAN DYSFUNCTION AT BASELINE AND 28 d AFTER ADMINISTRATION OF STUDY DRUG AND 28-d MORTALITY IN 247 PATIENTS WITH SEVERE SEPSIS OR EARLY SEPTIC SHOCK Placebo (n 5 78), % Organ Dysfunction Any Renal Pulmonary ARDS Non-ARDS Hepatic Hematologic CNS Hypotension 28-d all-cause mortality

Lenercept 0.042 mg/kg (n 5 82), %

Lenercept 0.083 mg/kg (n 5 87), %

Baseline

Day 28

Baseline

Day 28

Baseline

Day 28

(65) 83% (16) 21%

(16) 32% (4) 8%

(71) 87% (12) 15%

(12) 23% (0) 0%

(66) 76% (7) 8%

(13) 19% (0) 0%

(19) 24% (37) 47% (5) 6.4% (12) 15% (12) 15% (5) 6.4%

(6) 12% (7) 14% (2) 4% (1) 2% (2) 4% (3) 6%

(21) 26% (38) 46% (8) 9.8% (11) 13% (13) 16% (5) 6.1%

(5) 9.6% (5) 9.6% (3) 5.8% (2) 3.8% (2) 3.8% (1) 1.9%

(13) 15% (47) 54% (3) 3.4% (8) 9.2% (14) 16% (5) 5.7%

(1) 1.5% (11) 16% (0) 0% (1) 1.5% (1) 1.5% (0) 0%

28/78, 36%

30/82, 37%

20/87, 23%

854

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

odology (30). The 95% CI for the competing risk estimate of mean duration was computed using a non-parametric bootstrap method (31). All tests of significance were two-tailed. Mean values are given 6 1 standard deviation (SD) when specified or standard error (SE). Statistical analyses were carried out using SAS version 6.12 (Statistical Analysis System Institute, Cary, NC) and EpiInfo 6.0 (CDC, Atlanta, GA); p values less than 0.05 were considered significant.

RESULTS Overall Patient Population

A total of 498 patients with severe sepsis or late septic shock were enrolled into a randomized, placebo-controlled, multicenter clinical trial to determine the safety and efficacy of p55-IgG (lenercept) (20). Overall, there was no significant decrease in mortality between placebo or lenercept in all infused patients (20). A total of 251 patients (50%) had late septic shock, whereas 247 had severe sepsis or early septic shock and received placebo (n 5 78, 32%), or lenercept 0.042 mg/kg (n 5 82, 33%) or lenercept 0.083 mg/kg (n 5 87, 35%). Mortality rates and prevalence of organ dysfunction at baseline and 28 d after administration of the study drug in the three patient groups with severe sepsis or early septic shock are presented in Table 1. As shown, there was no statistically significant mortality benefit in any of these predetermined subgroups. Characteristics of the Study Population

Eighty-seven of the patients with severe sepsis or early septic shock received 0.083 mg/kg lenercept. In this prospectively defined subgroup, treatment was associated with a 36% reduction in 28-d mortality compared with a placebo group (n 5

TABLE 2 BASELINE CHARACTERISTICS OF 165 PATIENTS WITH SEVERE SEPSIS OR EARLY SEPTIC SHOCK ASSIGNED TO RECEIVE PLACEBO OR LENERCEPT (0.083 mg/kg) Characteristics Age, yr, median Sex, M/F ICU admission diagnosis Infectious conditions Gastrointestinal conditions* Trauma Postoperative conditions† Other medical conditions‡ Underlying conditions Liver cirrhosis Cancer Immunosuppressive state Recent surgery Microbiologically documented infections Gram-positive Gram-negative Fungal Mixed Predicted mortality at study entry 0–25 26–50 51–75 76–100

Placebo (n 5 78)

Lenercept (n 5 87 )

56 55/23

60 48/39

45 (58%) 14 (18%) 5 (6%) 8 (10%) 6 (7.8%)

50 (57%) 18 (21%) 7 (8%) 3 (3.5%) 8 (9.2%)

6 (8%) 3 (3.8%) 2 (2.5%) 18 (23%) 42/78 (54%) 14 (33%) 16 (38%) 2 (4.8%) 10 (24%)

10 (11%) 1 (1.2%) 1 (1.2%) 17 (20%) 59/87 (68%) 20 (34%) 21 (36%) 3 (5%) 15 (25%)

16 (21%) 33 (42%) 20 (26%) 9 (12%)

8 (9%) 33 (38%) 33 (38%) 13 (15%)

* Includes gastrointestinal dysfunction/perforation and surgery for gastrointestinal perforation. † Includes all surgeries except those for gastrointestinal perforation. ‡ Includes all nonoperative medical conditions and respiratory failure due to conditions other than pneumonia. § Predicted mortality was calculated by APACHE Medical Systems Inc. using the sepsis mortality prediction equation customized to the study protocol’s inclusion and exclusion criteria (28); expressed as percent mortality.

VOL 160

1999

78); mortality rates were 23% versus 36%, respectively (p 5 0.07). This reduction in 28-d mortality after treatment with 0.083 mg/kg lenercept was observed in patients with severe sepsis in the presence or in the absence of early septic shock (20). Data from all 165 patients with severe sepsis or early septic shock assigned to receive placebo (n 5 78) or 0.083 mg/kg lenercept (n 5 87) were included in the current analysis; none was lost to follow-up at Day 28. The two groups of patients were similar regarding age, gender, reason for ICU admission, underlying disease and major comorbidities, site of infection, type of microorganism recovered, and predicted mortality assessed by the customized APACHE III score (Table 2). Infections were microbiologically documented in two-thirds (101/ 165) of the patients. Leading sources of infection were the respiratory tract and the intra-abdominal space. Fifty-three of 165 patients (32%) had documented bloodstream infection (lenercept, 27; placebo, 25). Length of Hospital and ICU Stay

The mean (6 SE) length of hospital LOS after study entry was 16.9 6 0.9 d for patients in the lenercept group and 17.4 6 1.1 d in the placebo group; the difference between treatments was 0.5 6 1.4 d (95% CI 22.3 to 3.3). The competing risk estimate of mean LOS for patients discharged alive from the hospital was 21.0 6 0.7 d for the lenercept group and 23.9 6 0.6 d in the placebo group, with a difference between groups of 2.9 6 0.9 d (95% CI 1.1 to 4.7). Length of ICU stay averaged 9.8 6 0.7 d in the lenercept group and 12.4 6 1.0 d in the placebo group, with a difference of 2.6 6 1.2 d (95% CI 0.2 to 5.0). The competing risk estimate of the mean LOS among patients discharged alive from the ICU was 15.0 6 1.0 d for the lenercept group and 19.1 6 0.8 d in the placebo group, with a difference between groups of 4.1 6 1.3 d (95% CI 1.6 to 6.6). The estimated cumulative incidence of ICU discharge or all-cause mortality is shown in Figure 1. Organ Failures

The overall prevalence and distribution of organ dysfunction at study entry was similar in the two groups of patients (Table 1). New organ failures developed after study entry in approximately half of the patients (81 of 165, 49%). The proportion of patients with new organ failure was 43% in the lenercept group

Figure 1. Estimated cumulative incidence of ICU discharge or allcause mortality; Kaplan-Meier estimates, days to discharge from the ICU or death. Cumulative proportions of patients discharged from the ICU in the two groups of patients with severe sepsis or early septic shock treated with lenercept (solid line) or placebo (dashed line), respectively.

855

Pittet, Harbarth, Suter, et al.: Immunomodulating Therapy and Sepsis Morbidity TABLE 3 INCIDENCE-DENSITY OF END-ORGAN FAILURES IN 165 PATIENTS WITH SEVERE SEPSIS TREATED WITH PLACEBO (n 5 78) OR LENERCEPT (n 5 87) (ORGAN FAILURE DAYS* PER 1,000 PATIENT-DAYS) Total

†

Day 3 (74; 84) Day 7 (67; 79) Day 14 (58; 73) Day 21 (52; 67) Day 28 (50; 67)

ARDS

ARF

Hematologic

Hypotension

Placebo

Lenercept

Placebo

Lenercept

Placebo

Lenercept

Placebo

Lenercept

Placebo

Lenercept

1,579 1,254 724 481 500

1,310 848 452 239 209

223 186 134 110 81

195 133 67 56 47

177 150 93 77 70

134 89 75 37 17

154 97 31 33 47

168 59 17 28 9

446 212 103 99 47

382 207 108 28 9

Definition of abbreviation: ARF 5 acute renal failure. * Organ failure days have been summed up for each patient after study entry until Day 28, or discharge from the hospital or death when it occurred earlier. † Number of surviving patients in each study group (placebo; lenercept, respectively) is indicated in parentheses.

and 56% in the placebo group. The incidence of new organ failure was reduced in patients treated with lenercept compared with placebo: 41 versus 73 episodes per 1,000 patientdays, respectively (RR 0.57, 95% CI 0.29 to 1.10, p 5 0.10). Trends toward a reduction in the incidence of newly occurring organ failure were observed for liver failure (5% versus 15%, 67% decrease), DIC (6% versus 18%, 67% decrease), ARDS (9% versus 19%, 43% decrease), non-ARDS respiratory failure (23% versus 34%, 36% decrease), and hypotension (28% versus 40%, 37% decrease). The follow-up period (to Day 28, or death) for the 165 patients represented 3,592 patient-days. Patients in the cohort had a total of 2,488 organ failure days after study entry, with an average cumulative incidence across all patients of 693 per 1,000 patient-days. The overall incidence-density of organ failure was significantly lower in patients who received lenercept than in those who received placebo: 560 versus 858 organ failure days per 1,000 patient-days, respectively (RR 0.65; 95% CI 0.60 to 0.71, p 5 0.0001). Of note, incidence in the two groups increased 1.2-fold within the first 24 h after study entry, to reach maximal values at Day 3. The rate decreased by 24% between study entry and Day 7 in the lenercept group compared with only 8% in those who received placebo; between Days 3 and 7 it decreased by 35% (lenercept) compared with

A total of 116 patients (70%) presented with non-ARDS respiratory failure (n 5 84) or ARDS (n 5 32) at study entry. The mean duration of ventilatory support was 10.2 6 0.9 d for patients in the lenercept group and 10.8 6 1.0 d in the placebo group. The competing estimate of mean duration of ventilatory support among patients ultimately extubated and surviving was 14.2 6 1.0 d for the lenercept group and 17.4 6 1.3 d in the placebo group resulting in a difference between treatments of 3.2 6 1.6 d (95% CI 0.1 to 6.3). The cumulative incidence of extubation in the two groups of patients is shown in Figure 3. A total of 32 patients fulfilled the ARDS criteria at study entry (13 in the lenercept group and 19 in the placebo group);

Figure 2. Organ failure–free days. Bars indicate the total number of days during which patients in the two groups were free of each type of protocol-defined organ failure studied: renal (acute renal failure), respiratory ARDS, non-ARDS respiratory failure, hepatic, hematologic, CNS failure, and hypotension, respectively. Open columns: placebo; solid columns: lenercept.

Figure 3. Estimated cumulative incidence of extubation; KaplanMeier estimates, days to successful extubation. Cumulative proportions of patients withdrawn from ventilatory support in the two groups of patients with severe sepsis or early septic shock treated with lenercept (solid line) or placebo (dashed line), respectively.

only 21% with placebo (Table 3). By Day 21, the cumulative incidence-density of end-organ failure was reduced by 50% in the treatment group compared with placebo. There was a trend toward more days free of non-ARDS respiratory failure and ARDS, renal, hepatic, hematologic, and CNS failure (Figure 2) in the lenercept group, as well as toward more days free of organ system failure overall: 12,860 (lenercept) versus 9,796 (placebo), respectively. Ventilatory Support and ARDS

856

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

38% of these patients died. Among those who survived (9 of 13, 69% and 13 of 19, 68%, respectively), the number of days during which patients treated with lenercept met the criteria for ARDS tended to be lower in patients than in those treated with placebo (5.9 6 3.4 versus 14.1 6 3.5 d, p 5 0.088). The median LOS in ICU free of ARDS was 23 d (range 4 to 28) in lenercept-treated patients versus 17 d (range 0 to 27) in those who received placebo. Considering the entire cohort of patients who had ARDS at any point, there were a total of 432 d with ARDS, for an overall incidence of 120 ARDS-days per 1,000 patient-days of follow-up. The number of days with ARDS markedly differed between patients treated with lenercept (n 5 142 d, incidencedensity 5 71 ARDS-days per 1,000 patient-days) and those treated with placebo (n 5 290 d, incidence-density 5 182 ARDS-days per 1,000 patient-days; RR 5 0.39, 95% CI 0.32 to 0.48, p 5 0.0001). The incidence of ARDS was greatest on Day 3 (Table 3) and decreased between Days 3 and 7 by 32% (lenercept) versus 17% (placebo). By Day 21, the cumulative incidence of ARDS had reduced to 56 (lenercept) and 110 (placebo) ARDS-days per 1,000 patient-days.

DISCUSSION Treatment with lenercept (0.083 mg/kg) of prospectively defined patients with severe sepsis or early septic shock was associated with a trend toward reduced all-cause mortality (36% reduction) after 28 d of follow-up (20). The present study of the incidence and duration of organ failures was performed to test the hypothesis whether the decreased 28-d mortality was associated with a prolonged duration of intensive care to support failing organs, in particular among survivors (24, 32). In theory, resource use in ICU patients could be increased by effective new immunotherapies, unless there is a favorable impact on morbidity, resulting in a possible reduction in length of ICU and hospital stay. In contrast to the aforementioned hypothesis, this study suggests that in the patients with severe sepsis or early septic shock, therapy with lenercept is not associated with an increased ICU stay. Instead, the average ICU stay was 2.6 d shorter in the treatment group; for those patients who survived, this difference reached 4.1 d (95% CI 1.6 to 6.6). Importantly, therapy with lenercept was associated with a decreased incidence and duration of organ failures. As recently suggested by others (23, 26), the analysis of “days free of organ failure” was performed to obtain a more sensitive measure of effect than that provided by the analysis of mortality. This analysis is particularly important because of the confounding effect of early mortality on measures such as duration of organ failure, length of ICU stay, and ICU resource use (6, 26). Of importance, approximately half of the patients enrolled (49%) developed new organ failure after study entry. Both the number of patients with new organ failures and the duration of these dysfunctions were however decreased in patients treated with lenercept. The number of days with organ failure differed between the two groups of patients. As a consequence, organ failure–free days after study entry were more frequent in patients treated with lenercept compared with placebo. Both the incidence and duration of ARDS were decreased in the treatment group and withdrawal of ventilatory support was possible earlier (3.2 d, 95% CI 0.1 to 6.3) in treated survivors. Consequently, median length of ICU stay free of ARDS was markedly longer in patients treated with lenercept. Duration of ICU stay is strongly associated with increased risk for nosocomial infection (33, 34). These infections are one

VOL 160

1999

of the most important drivers for resource use and associated costs in the ICU, a point which is particularly true for ventilator-associated pneumonia and bloodstream infection (6, 34). Reducing length of ICU stay, and in particular reducing ventilator days and the use of invasive devices, would therefore be associated with reduced risk for infection and related extra costs. Noteworthy, only six episodes of ICU-acquired pneumonia developed after study entry in patients treated with lenercept (6 of 87, 7%) compared with 13 episodes (13 of 78, 17%) in those who received placebo (odds ratio 5 0.37, 95% CI 0.12 to 1.12, p 5 0.08). The current analysis has several limitations. First, it was conducted on a subset of patients recruited into a Phase II trial. However, we performed—as previously described (35)— a proper subgroup analysis of patients with characteristics defined before randomization and prospectively stratified into different study arms (20). Thus, we exclude a major bias by performing this analysis. Second, in a subsequent large Phase III trial in which mortality was the primary endpoint, lenercept failed to demonstrate a significant impact on survival of patients with severe sepsis or early septic shock, limiting the future application of the present study findings. Whether a similar impact on organ failures after immunomodulating therapy was also observed in the Phase III trial remains to be studied. Third, direct impact on patient morbidity or treatment costs arising from specific organ failures cannot be derived from our data. In particular, some organ failures are associated with a larger effect than others and organ dysfunctions may interact in many different ways. Similarly, the impact of nosocomial infections—a major cause of morbidity and mortality in noncoronary care ICUs, and resulting mainly from underlying disease and prolonged ICU stay, has not been emphasized in the present work. Fourth, length of ICU stay may vary between hospitals, physicians, and countries (33). Fifth, the dichotomization of patients as having specific organ failure or not using predefined cutoff values may not be the most accurate approach to analyze different organ dysfunction scenarios and should be reconsidered for future trials assessing patient morbidity and organ failure as important study endpoints. Finally, the data presented include only patients recruited within a clinical trial, and cannot be generalized to patients excluded from this study (20). The use of 28-d all-cause mortality as the standard endpoint for ICU-based and sepsis-related clinical trials assumes a homogeneous patient population and requires large sample sizes to show a possible benefit of therapy. Only few interventions currently used in critical care have unequivocally demonstrated their capacity to reduce mortality. Measures reflecting ICU-specific morbidity such as severity of organ dysfunction, nosocomial infection, or LOS are other relevant endpoints in the ICU setting. As stated by Petros and colleagues (26), regulatory agencies such as the Food and Drug Administration have recently begun to doubt the primacy of all-cause mortality as an endpoint for ICU-based clinical trials. Our findings suggest an association between the decreased incidence and duration of organ failure and the subsequent shorter length of ICU stay in patients with severe sepsis receiving immunomodulating therapy. Although the presented data do not prove causality, overall morbidity merits to be studied more actively in intensive care using aggregate variables with appropriate analytical techniques. Whether morbidity should replace mortality as an endpoint for clinical trials dealing with new sepsis therapies is beyond the scope of this discussion, but morbidity should at least be carefully monitored, appropriately studied and reported to improve our understanding of the disease process and impact.

Pittet, Harbarth, Suter, et al.: Immunomodulating Therapy and Sepsis Morbidity References 1. Fein, A. M., M. Lippmann, H. Holtzman, A. Eliraz, and S. K. Goldberg. 1983. The risk factors, incidence, and prognosis of ARDS following septicemia. Chest 83:40–42. 2. Parrillo, J. E., M. M. Parker, C. Natanson, A. F. Suffredini, R. L. Danner, R. E. Cunnion, and F. P. Ognibene. 1990. Septic shock in humans. Advances in the understanding of pathogenesis, cardiovascular dysfunction, and therapy. Ann. Intern. Med. 113:227–242. 3. Rangel-Frausto, S. M., D. Pittet, M. Costigan, T. S. Hwang, C. S. Davis, and R. P. Wenzel. 1995. The natural history of the systemic inflammatory response syndrome (SIRS): a prospective study. J.A.M.A. 273: 117–123. 4. Valles, J., C. Leon, and F. Alvarez-Lerma. 1997. Nosocomial bacteremia in critically ill patients: a multicenter study evaluating epidemiology and prognosis: Spanish collaborative group for infections in intensive care units. Clin. Infect. Dis. 24:387–395. 5. Pittet, D., B. Thiévent, R. P. Wenzel, N. Li, R. Auckenthaler, and P. M. Suter. 1996. Bedside prediction of mortality from bacteremic sepsis: a dynamic analysis of ICU patients. Am. J. Respir. Crit. Care Med. 153: 684–693. 6. Pittet, D., D. Tarara, and R. P. Wenzel. 1994. Nosocomial bloodstream infection in critically ill patients: excess length of stay, extra costs, and attributable mortality. J.A.M.A. 271:1598–1601. 7. van der Poll, T., H. R. Buller, H. ten Cate, C. H. Wortel, K. A. Bauer, S. J. van Deventer, C. E. Hack, H. P. Sauerwein, R. D. Rosenberg, and J. W. ten Cate. 1990. Activation of coagulation after administration of tumor necrosis factor to normal subjects. N. Engl. J. Med. 322: 1622–1627. 8. Leeper-Woodford, S. K., P. D. Carey, K. Byrne, J. K. Jenkins, B. J. Fisher, C. Blocher, H. J. Sugerman, and A. A. Fowler, 3d. 1991. Tumor necrosis factor. Am. Rev. Respir. Dis. 143(5, Pt. 1):1076–1082. 9. Damas, P., A. Reuter, P. Gysen, J. Demonty, M. Lamy, and P. Franchimont. 1989. Tumor necrosis factor and interleukin-1 serum levels during severe sepsis in humans. Crit. Care Med. 17:975–978. 10. Waage, A., A. Halstensen, and T. Espevik. 1987. Association between tumor necrosis factor in serum and fatal outcome in patients with meningococcal disease. Lancet 1:355–357. 11. Grau, G. E., T. E. Taylor, M. E. Molyneux, J. J. Wirima, P. Vassalli, M. Hommel, and P. H. Lambert. 1989. Tumor necrosis factor and disease severity in children with falciparum malaria. N. Engl. J. Med. 320:1586– 1591. 12. Kwiatkowski, D., A. V. Hill, I. Sambou, P. Twumasi, J. Castracane, K. R. Manogue, A. Cerami, D. R. Brewster, and B. M. Greenwood. 1990. TNF concentration in fatal cerebral, non-fatal cerebral, and uncomplicated Plasmodium falciparum malaria. Lancet 336:1201–1204. 13. Tracey, K. J., Y. Fong, and D. G. Hesse. 1987. Anti-cachectin/TNF monoclonal antibodies prevent septic shock during lethal bacteremia. Nature 330:662–664. 14. Opal, S. M., A. S. Cross, N. M. Kelly, J. C. Sadoff, M. W. Bodmer, J. E. Palardy, and G. H. Victor. 1990. Efficacy of a monoclonal antibody directed against tumor necrosis factor in protecting neutropenic rats from lethal infection with Pseudomonas aeruginosa. J. Infect. Dis. 161: 1148–1152. 15. Walsh, C. J., H. J. Sugerman, P. G. Mullen, P. D. Carey, S. K. LeeperWoodford, G. J. Jesmok, E. F. Ellis, and A. A. Fowler. 1992. Monoclonal antibody to tumor necrosis factor alpha attenuates cardiopulmonary dysfunction in porcine Gram-negative sepsis. Arch. Surg. 127: 138–144; Discussion 144–145. 16. Abraham, E., R. Wunderink, H. Silverman, T. M. Perl, S. Nasraway, H. Levy, R. Bone, R. P. Wenzel, R. Balk, R. Allred, and the TNF-alpha MAb Sepsis Study Group. 1995. Efficacy and safety of monoclonal antibody to human tumor necrosis factor alpha in patients with sepsis syndrome: a randomized, controlled, double-blind, multicenter clinical trial. J.A.M.A. 273:934–941. 17. Reinhart, K., C. Wiegand-Lohnert, F. Grimminger, M. Kaul, S. Withington, D. Treacher, J. Eckart, S. Willatts, C. Bouza, D. Krausch, F. Stockenhuber, J. Eiselstein, L. Daum, and J. Kempeni. 1996. Assessment of the safety and efficacy of the monoclonal anti-tumor necrosis

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29. 30. 31. 32.

33.

34.

35.

857

factor antibody-fragment, MAK 195F, in patients with sepsis and septic shock. Crit. Care Med. 24:733–742. Cohen, J., and J. Carlet. 1996. INTERSEPT: an international, multicenter, placebo-controlled trial of monoclonal antibody to human tumor necrosis factor-alpha in patients with sepsis. International Sepsis Trial Study Group. Crit. Care Med. 24:1431–1440. Fisher, C. J., Jr., J. M. Agosti, S. M. Opal, S. F. Lowry, R. A. Balk, J. C. Sadoff, E. Abraham, R. M. Schein, and E. Benjamin. 1996. Treatment of septic shock with the tumor necrosis factor receptor:Fc fusion protein. N. Engl. J. Med. 334:1697–1702. Abraham, E., M. P. Glauser, T. Butler, J. Garbino, D. Gelmont, P. F. Laterre, K. Kudsk, H. A. Bruining, C. Otto, E. Tobin, and the RO 452081 Study Group. 1997. P55 tumor necrosis factor receptor fusion protein in the treatment of patients with severe sepsis and septic shock: a randomized controlled multicenter trial. J.A.M.A. 277:1531– 1538. Hinshaw, L. B., T. E. Emerson, Jr., F. B. Taylor, Jr., A. C. Chang, M. Duerr, G. T. Peer, D. J. Flournoy, G. L. White, S. D. Kosanke, and C. K. Murray. 1992. Lethal Staphylococcus aureus–induced shock in primates: prevention of death with anti-TNF antibody. J. Trauma 33: 568–573. Jin, H., R. Yang, S. A. Marsters, S. A. Bunting, F. M. Wurm, S. M. Chamow, and A. Ashkenazi. 1994. Protection against rat endotoxic shock by p55 tumor necrosis factor (TNF) receptor immunoadhesin: comparison with anti-TNF monoclonal antibody. J. Infect. Dis. 170:1323–1326. Bernard, G. R., A. P. Wheeler, J. A. Russell, R. Schein, W. R. Summer, K. P. Steinberg, W. J. Fulkerson, P. E. Wright, B. W. Christman, W. D. Dupont, S. B. Higgins, B. B. Swindell, and the Ibuprofen in Sepsis Study Group. 1997. The effects of ibuprofen on the physiology and survival of patients with sepsis. N. Engl. J. Med. 336:912–918. Schulman, K. A., H. A. Glick, H. Rubin, and J. M. Eisenberg. 1991. Cost-effectiveness of HA-1A monoclonal antibody for Gram-negative sepsis: economic assessment of a new therapeutic agent. J.A.M.A. 266: 3466–3471. Chalfin, D. B., M. E. Holbein, A. M. Fein, and G. C. Carlon. 1993. Costeffectiveness of monoclonal antibodies to Gram-negative endotoxin in the treatment of Gram-negative sepsis in ICU patients. J.A.M.A. 269:249–254. Petros, A. J., J. C. Marshall, and H. K. F. Van Saene. 1995. Should morbidity replace mortality as an endpoint for clinical trials in intensive care? Lancet 345:369–371. Hebert, P. C., A. J. Drummond, J. Singer, G. R. Bernard, and J. A. Russell. 1993. A simple multiple system organ failure scoring system predicts mortality of patients who have sepsis syndrome. Chest 104:230–235. Knaus, W. A., F. E. Harrell, C. J. Fisher, D. P. Wagner, S. M. Opal, J. C. Sadoff, E. A. Draper, C. A. Walawander, K. Conboy, and T. H. Grasela. 1993. The clinical evaluation of new drugs for sepsis: a prospective study design based on survival analysis. J.A.M.A. 270:1233–1241. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete information. J. Am. Stat. Assoc. 53:457–481. Pepe, M. S. 1991. Inference for events with dependent risks in multiple endpoint studies. J. Am. Stat. Assoc. 86:770–778. Efron, B., and R. Tibshirani, editors. 1993. An Introduction to the Bootstrap. Chapman and Hall, New York. Barriere, S. L., and B. J. Guglielmo. 1992. Gram-negative sepsis, the sepsis syndrome, and the role of antiendotoxin monoclonal antibodies. Clin. Pharm. 11:223–235. Vincent, J. L., D. J. Bihari, P. M. Suter, H. A. Bruining, J. White, M.-H. Nicolas-Chanoin, M. Wolff, R. C. Spencer, and M. Hemmer. 1995. The prevalence of nosocomial infection in intensive care units in Europe: results of the European Prevalence of Infection in Intensive Care (EPIC) Study. J.A.M.A. 274:639–644. Pittet, D., and S. Harbarth. 1998. The intensive care unit. In J. V. Bennett and P. S. Brachman, editors. Hospital Infections, 4th ed. Little, Brown, Boston. 381–402. Yusuf, S., J. Wittes, J. Probstfield, and H. A. Tyroler. 1991. Analysis and interpretation of treatment effects in subgroups of patients in randomized clinical trials. J.A.M.A. 266:93–98.